ViewVC Help

View File | Revision Log | Show Annotations | Download File

/cvs/AnyEvent-HTTP/HTTP.pm

Comparing AnyEvent-HTTP/HTTP.pm (file contents):
Revision 1.2 by root, Wed Jun 4 11:37:41 2008 UTC vs.
Revision 1.138 by root, Fri Aug 5 20:45:09 2022 UTC

…		…
3	AnyEvent::HTTP - simple but non-blocking HTTP/HTTPS client	3	AnyEvent::HTTP - simple but non-blocking HTTP/HTTPS client
4		4
5	=head1 SYNOPSIS	5	=head1 SYNOPSIS
6		6
7	use AnyEvent::HTTP;	7	use AnyEvent::HTTP;
		8
		9	http_get "http://www.nethype.de/", sub { print $_[1] };
		10
		11	# ... do something else here
8		12
9	=head1 DESCRIPTION	13	=head1 DESCRIPTION
10		14
11	This module is an L<AnyEvent> user, you need to make sure that you use and	15	This module is an L<AnyEvent> user, you need to make sure that you use and
12	run a supported event loop.	16	run a supported event loop.
13		17
		18	This module implements a simple, stateless and non-blocking HTTP
		19	client. It supports GET, POST and other request methods, cookies and more,
		20	all on a very low level. It can follow redirects, supports proxies, and
		21	automatically limits the number of connections to the values specified in
		22	the RFC.
		23
		24	It should generally be a "good client" that is enough for most HTTP
		25	tasks. Simple tasks should be simple, but complex tasks should still be
		26	possible as the user retains control over request and response headers.
		27
		28	The caller is responsible for authentication management, cookies (if
		29	the simplistic implementation in this module doesn't suffice), referer
		30	and other high-level protocol details for which this module offers only
		31	limited support.
		32
14	=head2 METHODS	33	=head2 METHODS
15		34
16	=over 4	35	=over 4
17		36
18	=cut	37	=cut
19		38
20	package AnyEvent::HTTP;	39	package AnyEvent::HTTP;
21		40
22	use strict;	41	use common::sense;
23	no warnings;
24		42
25	use Carp;	43	use Errno ();
26		44
27	use AnyEvent ();	45	use AnyEvent 5.0 ();
28	use AnyEvent::Util ();	46	use AnyEvent::Util ();
29	use AnyEvent::Socket ();
30	use AnyEvent::Handle ();	47	use AnyEvent::Handle ();
31		48
32	use base Exporter::;	49	use base Exporter::;
33		50
34	our $VERSION = '1.0';	51	our $VERSION = 2.25;
35		52
36	our @EXPORT = qw(http_get http_request);	53	our @EXPORT = qw(http_get http_post http_head http_request);
37		54
38	our $USERAGENT = "Mozilla/5.0 (compatible; AnyEvent::HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)";	55	our $USERAGENT = "Mozilla/5.0 (compatible; U; AnyEvent-HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)";
39	our $MAX_REDIRECTS = 10;	56	our $MAX_RECURSE = 10;
40	our $MAX_PERSISTENT = 8;
41	our $PERSISTENT_TIMEOUT = 2;	57	our $PERSISTENT_TIMEOUT = 3;
42	our $TIMEOUT = 300;	58	our $TIMEOUT = 300;
43		59	our $MAX_PER_HOST = 4; # changing this is evil
44	# changing these is evil
45	our $MAX_PERSISTENT_PER_HOST = 2;
46	our $MAX_PER_HOST = 4; # not respected yet :(
47		60
48	our $PROXY;	61	our $PROXY;
		62	our $ACTIVE = 0;
49		63
50	my %KA_COUNT; # number of open keep-alive connections per host	64	my %KA_CACHE; # indexed by uhost currently, points to [$handle...] array
		65	my %CO_SLOT; # number of open connections, and wait queue, per host
51		66
52	=item http_get $url, key => value..., $cb->($data, $headers)	67	=item http_get $url, key => value..., $cb->($data, $headers)
53		68
54	Executes an HTTP-GET request. See the http_request function for details on	69	Executes an HTTP-GET request. See the http_request function for details on
55	additional parameters.	70	additional parameters and the return value.
		71
		72	=item http_head $url, key => value..., $cb->($data, $headers)
		73
		74	Executes an HTTP-HEAD request. See the http_request function for details
		75	on additional parameters and the return value.
		76
		77	=item http_post $url, $body, key => value..., $cb->($data, $headers)
		78
		79	Executes an HTTP-POST request with a request body of C<$body>. See the
		80	http_request function for details on additional parameters and the return
		81	value.
56		82
57	=item http_request $method => $url, key => value..., $cb->($data, $headers)	83	=item http_request $method => $url, key => value..., $cb->($data, $headers)
58		84
59	Executes a HTTP request of type C<$method> (e.g. C<GET>, C<POST>). The URL	85	Executes a HTTP request of type C<$method> (e.g. C<GET>, C<POST>). The URL
60	must be an absolute http or https URL.	86	must be an absolute http or https URL.
61		87
		88	When called in void context, nothing is returned. In other contexts,
		89	C<http_request> returns a "cancellation guard" - you have to keep the
		90	object at least alive until the callback get called. If the object gets
		91	destroyed before the callback is called, the request will be cancelled.
		92
62	The callback will be called with the response data as first argument	93	The callback will be called with the response body data as first argument
63	(or C<undef> if it wasn't available due to errors), and a hash-ref with	94	(or C<undef> if an error occurred), and a hash-ref with response headers
64	response headers as second argument.	95	(and trailers) as second argument.
65		96
66	All the headers in that has are lowercased. In addition to the response	97	All the headers in that hash are lowercased. In addition to the response
67	headers, the three "pseudo-headers" C<HTTPVersion>, C<Status> and	98	headers, the "pseudo-headers" (uppercase to avoid clashing with possible
68	C<Reason> contain the three parts of the HTTP Status-Line of the same	99	response headers) C<HTTPVersion>, C<Status> and C<Reason> contain the
69	name.	100	three parts of the HTTP Status-Line of the same name. If an error occurs
		101	during the body phase of a request, then the original C<Status> and
		102	C<Reason> values from the header are available as C<OrigStatus> and
		103	C<OrigReason>.
		104
		105	The pseudo-header C<URL> contains the actual URL (which can differ from
		106	the requested URL when following redirects - for example, you might get
		107	an error that your URL scheme is not supported even though your URL is a
		108	valid http URL because it redirected to an ftp URL, in which case you can
		109	look at the URL pseudo header).
		110
		111	The pseudo-header C<Redirect> only exists when the request was a result
		112	of an internal redirect. In that case it is an array reference with
		113	the C<($data, $headers)> from the redirect response. Note that this
		114	response could in turn be the result of a redirect itself, and C<<
		115	$headers->{Redirect}[1]{Redirect} >> will then contain the original
		116	response, and so on.
		117
		118	If the server sends a header multiple times, then their contents will be
		119	joined together with a comma (C<,>), as per the HTTP spec.
70		120
71	If an internal error occurs, such as not being able to resolve a hostname,	121	If an internal error occurs, such as not being able to resolve a hostname,
72	then C<$data> will be C<undef>, C<< $headers->{Status} >> will be C<599>	122	then C<$data> will be C<undef>, C<< $headers->{Status} >> will be
73	and the C<Reason> pseudo-header will contain an error message.	123	C<590>-C<599> and the C<Reason> pseudo-header will contain an error
		124	message. Currently the following status codes are used:
		125
		126	=over 4
		127
		128	=item 595 - errors during connection establishment, proxy handshake.
		129
		130	=item 596 - errors during TLS negotiation, request sending and header processing.
		131
		132	=item 597 - errors during body receiving or processing.
		133
		134	=item 598 - user aborted request via C<on_header> or C<on_body>.
		135
		136	=item 599 - other, usually nonretryable, errors (garbled URL etc.).
		137
		138	=back
		139
		140	A typical callback might look like this:
		141
		142	sub {
		143	my ($body, $hdr) = @_;
		144
		145	if ($hdr->{Status} =~ /^2/) {
		146	... everything should be ok
		147	} else {
		148	print "error, $hdr->{Status} $hdr->{Reason}\n";
		149	}
		150	}
74		151
75	Additional parameters are key-value pairs, and are fully optional. They	152	Additional parameters are key-value pairs, and are fully optional. They
76	include:	153	include:
77		154
78	=over 4	155	=over 4
79		156
80	=item recurse => $boolean (default: true)	157	=item recurse => $count (default: $MAX_RECURSE)
81		158
82	Whether to recurse requests or not, e.g. on redirects, authentication	159	Whether to recurse requests or not, e.g. on redirects, authentication and
83	retries and so on.	160	other retries and so on, and how often to do so.
		161
		162	Only redirects to http and https URLs are supported. While most common
		163	redirection forms are handled entirely within this module, some require
		164	the use of the optional L<URI> module. If it is required but missing, then
		165	the request will fail with an error.
84		166
85	=item headers => hashref	167	=item headers => hashref
86		168
87	The request headers to use.	169	The request headers to use. Currently, C<http_request> may provide its own
		170	C<Host:>, C<Content-Length:>, C<Connection:> and C<Cookie:> headers and
		171	will provide defaults at least for C<TE:>, C<Referer:> and C<User-Agent:>
		172	(this can be suppressed by using C<undef> for these headers in which case
		173	they won't be sent at all).
		174
		175	You really should provide your own C<User-Agent:> header value that is
		176	appropriate for your program - I wouldn't be surprised if the default
		177	AnyEvent string gets blocked by webservers sooner or later.
		178
		179	Also, make sure that your headers names and values do not contain any
		180	embedded newlines.
88		181
89	=item timeout => $seconds	182	=item timeout => $seconds
90		183
91	The time-out to use for various stages - each connect attempt will reset	184	The time-out to use for various stages - each connect attempt will reset
92	the timeout, as will read or write activity. Default timeout is 5 minutes.	185	the timeout, as will read or write activity, i.e. this is not an overall
		186	timeout.
		187
		188	Default timeout is 5 minutes.
93		189
94	=item proxy => [$host, $port[, $scheme]] or undef	190	=item proxy => [$host, $port[, $scheme]] or undef
95		191
96	Use the given http proxy for all requests. If not specified, then the	192	Use the given http proxy for all requests, or no proxy if C<undef> is
97	default proxy (as specified by C<$ENV{http_proxy}>) is used.	193	used.
98		194
99	C<$scheme> must be either missing or C<http> for HTTP, or C<https> for	195	C<$scheme> must be either missing or must be C<http> for HTTP.
100	HTTPS.	196
		197	If not specified, then the default proxy is used (see
		198	C<AnyEvent::HTTP::set_proxy>).
		199
		200	Currently, if your proxy requires authorization, you have to specify an
		201	appropriate "Proxy-Authorization" header in every request.
		202
		203	Note that this module will prefer an existing persistent connection,
		204	even if that connection was made using another proxy. If you need to
		205	ensure that a new connection is made in this case, you can either force
		206	C<persistent> to false or e.g. use the proxy address in your C<sessionid>.
		207
		208	=item body => $string
		209
		210	The request body, usually empty. Will be sent as-is (future versions of
		211	this module might offer more options).
		212
		213	=item cookie_jar => $hash_ref
		214
		215	Passing this parameter enables (simplified) cookie-processing, loosely
		216	based on the original netscape specification.
		217
		218	The C<$hash_ref> must be an (initially empty) hash reference which
		219	will get updated automatically. It is possible to save the cookie jar
		220	to persistent storage with something like JSON or Storable - see the
		221	C<AnyEvent::HTTP::cookie_jar_expire> function if you wish to remove
		222	expired or session-only cookies, and also for documentation on the format
		223	of the cookie jar.
		224
		225	Note that this cookie implementation is not meant to be complete. If
		226	you want complete cookie management you have to do that on your
		227	own. C<cookie_jar> is meant as a quick fix to get most cookie-using sites
		228	working. Cookies are a privacy disaster, do not use them unless required
		229	to.
		230
		231	When cookie processing is enabled, the C<Cookie:> and C<Set-Cookie:>
		232	headers will be set and handled by this module, otherwise they will be
		233	left untouched.
		234
		235	=item tls_ctx => $scheme | $tls_ctx
		236
		237	Specifies the AnyEvent::TLS context to be used for https connections. This
		238	parameter follows the same rules as the C<tls_ctx> parameter to
		239	L<AnyEvent::Handle>, but additionally, the two strings C<low> or
		240	C<high> can be specified, which give you a predefined low-security (no
		241	verification, highest compatibility) and high-security (CA and common-name
		242	verification) TLS context.
		243
		244	The default for this option is C<low>, which could be interpreted as "give
		245	me the page, no matter what".
		246
		247	See also the C<sessionid> parameter.
		248
		249	=item sessionid => $string
		250
		251	The module might reuse connections to the same host internally (regardless
		252	of other settings, such as C<tcp_connect> or C<proxy>). Sometimes (e.g.
		253	when using TLS or a specfic proxy), you do not want to reuse connections
		254	from other sessions. This can be achieved by setting this parameter to
		255	some unique ID (such as the address of an object storing your state data
		256	or the TLS context, or the proxy IP) - only connections using the same
		257	unique ID will be reused.
		258
		259	=item on_prepare => $callback->($fh)
		260
		261	In rare cases you need to "tune" the socket before it is used to
		262	connect (for example, to bind it on a given IP address). This parameter
		263	overrides the prepare callback passed to C<AnyEvent::Socket::tcp_connect>
		264	and behaves exactly the same way (e.g. it has to provide a
		265	timeout). See the description for the C<$prepare_cb> argument of
		266	C<AnyEvent::Socket::tcp_connect> for details.
		267
		268	=item tcp_connect => $callback->($host, $service, $connect_cb, $prepare_cb)
		269
		270	In even rarer cases you want total control over how AnyEvent::HTTP
		271	establishes connections. Normally it uses L<AnyEvent::Socket::tcp_connect>
		272	to do this, but you can provide your own C<tcp_connect> function -
		273	obviously, it has to follow the same calling conventions, except that it
		274	may always return a connection guard object.
		275
		276	The connections made by this hook will be treated as equivalent to
		277	connections made the built-in way, specifically, they will be put into
		278	and taken from the persistent connection cache. If your C<$tcp_connect>
		279	function is incompatible with this kind of re-use, consider switching off
		280	C<persistent> connections and/or providing a C<sessionid> identifier.
		281
		282	There are probably lots of weird uses for this function, starting from
		283	tracing the hosts C<http_request> actually tries to connect, to (inexact
		284	but fast) host => IP address caching or even socks protocol support.
		285
		286	=item on_header => $callback->($headers)
		287
		288	When specified, this callback will be called with the header hash as soon
		289	as headers have been successfully received from the remote server (not on
		290	locally-generated errors).
		291
		292	It has to return either true (in which case AnyEvent::HTTP will continue),
		293	or false, in which case AnyEvent::HTTP will cancel the download (and call
		294	the finish callback with an error code of C<598>).
		295
		296	This callback is useful, among other things, to quickly reject unwanted
		297	content, which, if it is supposed to be rare, can be faster than first
		298	doing a C<HEAD> request.
		299
		300	The downside is that cancelling the request makes it impossible to re-use
		301	the connection. Also, the C<on_header> callback will not receive any
		302	trailer (headers sent after the response body).
		303
		304	Example: cancel the request unless the content-type is "text/html".
		305
		306	on_header => sub {
		307	$_[0]{"content-type"} =~ /^text\/html\s*(?:;|$)/
		308	},
		309
		310	=item on_body => $callback->($partial_body, $headers)
		311
		312	When specified, all body data will be passed to this callback instead of
		313	to the completion callback. The completion callback will get the empty
		314	string instead of the body data.
		315
		316	It has to return either true (in which case AnyEvent::HTTP will continue),
		317	or false, in which case AnyEvent::HTTP will cancel the download (and call
		318	the completion callback with an error code of C<598>).
		319
		320	The downside to cancelling the request is that it makes it impossible to
		321	re-use the connection.
		322
		323	This callback is useful when the data is too large to be held in memory
		324	(so the callback writes it to a file) or when only some information should
		325	be extracted, or when the body should be processed incrementally.
		326
		327	It is usually preferred over doing your own body handling via
		328	C<want_body_handle>, but in case of streaming APIs, where HTTP is
		329	only used to create a connection, C<want_body_handle> is the better
		330	alternative, as it allows you to install your own event handler, reducing
		331	resource usage.
		332
		333	=item want_body_handle => $enable
		334
		335	When enabled (default is disabled), the behaviour of AnyEvent::HTTP
		336	changes considerably: after parsing the headers, and instead of
		337	downloading the body (if any), the completion callback will be
		338	called. Instead of the C<$body> argument containing the body data, the
		339	callback will receive the L<AnyEvent::Handle> object associated with the
		340	connection. In error cases, C<undef> will be passed. When there is no body
		341	(e.g. status C<304>), the empty string will be passed.
		342
		343	The handle object might or might not be in TLS mode, might be connected
		344	to a proxy, be a persistent connection, use chunked transfer encoding
		345	etc., and configured in unspecified ways. The user is responsible for this
		346	handle (it will not be used by this module anymore).
		347
		348	This is useful with some push-type services, where, after the initial
		349	headers, an interactive protocol is used (typical example would be the
		350	push-style twitter API which starts a JSON/XML stream).
		351
		352	If you think you need this, first have a look at C<on_body>, to see if
		353	that doesn't solve your problem in a better way.
		354
		355	=item persistent => $boolean
		356
		357	Try to create/reuse a persistent connection. When this flag is set
		358	(default: true for idempotent requests, false for all others), then
		359	C<http_request> tries to re-use an existing (previously-created)
		360	persistent connection to same host (i.e. identical URL scheme, hostname,
		361	port and sessionid) and, failing that, tries to create a new one.
		362
		363	Requests failing in certain ways will be automatically retried once, which
		364	is dangerous for non-idempotent requests, which is why it defaults to off
		365	for them. The reason for this is because the bozos who designed HTTP/1.1
		366	made it impossible to distinguish between a fatal error and a normal
		367	connection timeout, so you never know whether there was a problem with
		368	your request or not.
		369
		370	When reusing an existent connection, many parameters (such as TLS context)
		371	will be ignored. See the C<sessionid> parameter for a workaround.
		372
		373	=item keepalive => $boolean
		374
		375	Only used when C<persistent> is also true. This parameter decides whether
		376	C<http_request> tries to handshake a HTTP/1.0-style keep-alive connection
		377	(as opposed to only a HTTP/1.1 persistent connection).
		378
		379	The default is true, except when using a proxy, in which case it defaults
		380	to false, as HTTP/1.0 proxies cannot support this in a meaningful way.
		381
		382	=item handle_params => { key => value ... }
		383
		384	The key-value pairs in this hash will be passed to any L<AnyEvent::Handle>
		385	constructor that is called - not all requests will create a handle, and
		386	sometimes more than one is created, so this parameter is only good for
		387	setting hints.
		388
		389	Example: set the maximum read size to 4096, to potentially conserve memory
		390	at the cost of speed.
		391
		392	handle_params => {
		393	max_read_size => 4096,
		394	},
101		395
102	=back	396	=back
103		397
104	=back	398	Example: do a simple HTTP GET request for http://www.nethype.de/ and print
		399	the response body.
		400
		401	http_request GET => "http://www.nethype.de/", sub {
		402	my ($body, $hdr) = @_;
		403	print "$body\n";
		404	};
		405
		406	Example: do a HTTP HEAD request on https://www.google.com/, use a
		407	timeout of 30 seconds.
		408
		409	http_request
		410	HEAD => "https://www.google.com",
		411	headers => { "user-agent" => "MySearchClient 1.0" },
		412	timeout => 30,
		413	sub {
		414	my ($body, $hdr) = @_;
		415	use Data::Dumper;
		416	print Dumper $hdr;
		417	}
		418	;
		419
		420	Example: do another simple HTTP GET request, but immediately try to
		421	cancel it.
		422
		423	my $request = http_request GET => "http://www.nethype.de/", sub {
		424	my ($body, $hdr) = @_;
		425	print "$body\n";
		426	};
		427
		428	undef $request;
105		429
106	=cut	430	=cut
107		431
		432	#############################################################################
		433	# wait queue/slots
		434
		435	sub _slot_schedule;
		436	sub _slot_schedule($) {
		437	my $host = shift;
		438
		439	while ($CO_SLOT{$host}[0] < $MAX_PER_HOST) {
		440	if (my $cb = shift @{ $CO_SLOT{$host}[1] }) {
		441	# somebody wants that slot
		442	++$CO_SLOT{$host}[0];
		443	++$ACTIVE;
		444
		445	$cb->(AnyEvent::Util::guard {
		446	--$ACTIVE;
		447	--$CO_SLOT{$host}[0];
		448	_slot_schedule $host;
		449	});
		450	} else {
		451	# nobody wants the slot, maybe we can forget about it
		452	delete $CO_SLOT{$host} unless $CO_SLOT{$host}[0];
		453	last;
		454	}
		455	}
		456	}
		457
		458	# wait for a free slot on host, call callback
		459	sub _get_slot($$) {
		460	push @{ $CO_SLOT{$_[0]}[1] }, $_[1];
		461
		462	_slot_schedule $_[0];
		463	}
		464
		465	#############################################################################
		466	# cookie handling
		467
		468	# expire cookies
		469	sub cookie_jar_expire($;$) {
		470	my ($jar, $session_end) = @_;
		471
		472	%$jar = () if $jar->{version} != 2;
		473
		474	my $anow = AE::now;
		475
		476	while (my ($chost, $paths) = each %$jar) {
		477	next unless ref $paths;
		478
		479	while (my ($cpath, $cookies) = each %$paths) {
		480	while (my ($cookie, $kv) = each %$cookies) {
		481	if (exists $kv->{_expires}) {
		482	delete $cookies->{$cookie}
		483	if $anow > $kv->{_expires};
		484	} elsif ($session_end) {
		485	delete $cookies->{$cookie};
		486	}
		487	}
		488
		489	delete $paths->{$cpath}
		490	unless %$cookies;
		491	}
		492
		493	delete $jar->{$chost}
		494	unless %$paths;
		495	}
		496	}
		497
		498	# extract cookies from jar
		499	sub cookie_jar_extract($$$$) {
		500	my ($jar, $scheme, $host, $path) = @_;
		501
		502	%$jar = () if $jar->{version} != 2;
		503
		504	$host = AnyEvent::Util::idn_to_ascii $host
		505	if $host =~ /[^\x00-\x7f]/;
		506
		507	my @cookies;
		508
		509	while (my ($chost, $paths) = each %$jar) {
		510	next unless ref $paths;
		511
		512	# exact match or suffix including . match
		513	$chost eq $host or ".$chost" eq substr $host, -1 - length $chost
		514	or next;
		515
		516	while (my ($cpath, $cookies) = each %$paths) {
		517	next unless $cpath eq substr $path, 0, length $cpath;
		518
		519	while (my ($cookie, $kv) = each %$cookies) {
		520	next if $scheme ne "https" && exists $kv->{secure};
		521
		522	if (exists $kv->{_expires} and AE::now > $kv->{_expires}) {
		523	delete $cookies->{$cookie};
		524	next;
		525	}
		526
		527	my $value = $kv->{value};
		528
		529	if ($value =~ /[=;,[:space:]]/) {
		530	$value =~ s/([\\"])/\\$1/g;
		531	$value = "\"$value\"";
		532	}
		533
		534	push @cookies, "$cookie=$value";
		535	}
		536	}
		537	}
		538
		539	\@cookies
		540	}
		541
		542	# parse set_cookie header into jar
		543	sub cookie_jar_set_cookie($$$$) {
		544	my ($jar, $set_cookie, $host, $date) = @_;
		545
		546	%$jar = () if $jar->{version} != 2;
		547
		548	my $anow = int AE::now;
		549	my $snow; # server-now
		550
		551	for ($set_cookie) {
		552	# parse NAME=VALUE
		553	my @kv;
		554
		555	# expires is not http-compliant in the original cookie-spec,
		556	# we support the official date format and some extensions
		557	while (
		558	m{
		559	\G\s*
		560	(?:
		561	expires \s*=\s* ([A-Z][a-z][a-z]+,\ [^,;]+)
		562	| ([^=;,[:space:]]+) (?: \s*=\s* (?: "((?:[^\\"]+|\\.)*)" | ([^;,[:space:]]*) ) )?
		563	)
		564	}gcxsi
		565	) {
		566	my $name = $2;
		567	my $value = $4;
		568
		569	if (defined $1) {
		570	# expires
		571	$name = "expires";
		572	$value = $1;
		573	} elsif (defined $3) {
		574	# quoted
		575	$value = $3;
		576	$value =~ s/\\(.)/$1/gs;
		577	}
		578
		579	push @kv, @kv ? lc $name : $name, $value;
		580
		581	last unless /\G\s*;/gc;
		582	}
		583
		584	last unless @kv;
		585
		586	my $name = shift @kv;
		587	my %kv = (value => shift @kv, @kv);
		588
		589	if (exists $kv{"max-age"}) {
		590	$kv{_expires} = $anow + delete $kv{"max-age"};
		591	} elsif (exists $kv{expires}) {
		592	$snow ||= parse_date ($date) || $anow;
		593	$kv{_expires} = $anow + (parse_date (delete $kv{expires}) - $snow);
		594	} else {
		595	delete $kv{_expires};
		596	}
		597
		598	my $cdom;
		599	my $cpath = (delete $kv{path}) || "/";
		600
		601	if (exists $kv{domain}) {
		602	$cdom = $kv{domain};
		603
		604	$cdom =~ s/^\.?/./; # make sure it starts with a "."
		605
		606	next if $cdom =~ /\.$/;
		607
		608	# this is not rfc-like and not netscape-like. go figure.
		609	my $ndots = $cdom =~ y/.//;
		610	next if $ndots < ($cdom =~ /\.[^.][^.]\.[^.][^.]$/ ? 3 : 2);
		611
		612	$cdom = substr $cdom, 1; # remove initial .
		613	} else {
		614	$cdom = $host;
		615	}
		616
		617	# store it
		618	$jar->{version} = 2;
		619	$jar->{lc $cdom}{$cpath}{$name} = \%kv;
		620
		621	redo if /\G\s*,/gc;
		622	}
		623	}
		624
		625	#############################################################################
		626	# keepalive/persistent connection cache
		627
		628	# fetch a connection from the keepalive cache
		629	sub ka_fetch($) {
		630	my $ka_key = shift;
		631
		632	my $hdl = pop @{ $KA_CACHE{$ka_key} }; # currently we reuse the MOST RECENTLY USED connection
		633	delete $KA_CACHE{$ka_key}
		634	unless @{ $KA_CACHE{$ka_key} };
		635
		636	$hdl
		637	}
		638
		639	sub ka_store($$) {
		640	my ($ka_key, $hdl) = @_;
		641
		642	my $kaa = $KA_CACHE{$ka_key} ||= [];
		643
		644	my $destroy = sub {
		645	my @ka = grep $_ != $hdl, @{ $KA_CACHE{$ka_key} };
		646
		647	$hdl->destroy;
		648
		649	@ka
		650	? $KA_CACHE{$ka_key} = \@ka
		651	: delete $KA_CACHE{$ka_key};
		652	};
		653
		654	# on error etc., destroy
		655	$hdl->on_error ($destroy);
		656	$hdl->on_eof ($destroy);
		657	$hdl->on_read ($destroy);
		658	$hdl->timeout ($PERSISTENT_TIMEOUT);
		659
		660	push @$kaa, $hdl;
		661	shift @$kaa while @$kaa > $MAX_PER_HOST;
		662	}
		663
		664	#############################################################################
		665	# utilities
		666
		667	# continue to parse $_ for headers and place them into the arg
		668	sub _parse_hdr() {
		669	my %hdr;
		670
		671	# things seen, not parsed:
		672	# p3pP="NON CUR OTPi OUR NOR UNI"
		673
		674	$hdr{lc $1} .= ",$2"
		675	while /\G
		676	([^:\000-\037]*):
		677	[\011\040]*
		678	((?: [^\012]+ | \012[\011\040] )*)
		679	\012
		680	/gxc;
		681
		682	/\G$/
		683	or return;
		684
		685	# remove the "," prefix we added to all headers above
		686	substr $_, 0, 1, ""
		687	for values %hdr;
		688
		689	\%hdr
		690	}
		691
		692	#############################################################################
		693	# http_get
		694
		695	our $qr_nlnl = qr{(?<![^\012])\015?\012};
		696
		697	our $TLS_CTX_LOW = { cache => 1, sslv2 => 1 };
		698	our $TLS_CTX_HIGH = { cache => 1, verify => 1, verify_peername => "https" };
		699
		700	# maybe it should just become a normal object :/
		701
		702	sub _destroy_state(\%) {
		703	my ($state) = @_;
		704
		705	$state->{handle}->destroy if $state->{handle};
		706	%$state = ();
		707	}
		708
		709	sub _error(\%$$) {
		710	my ($state, $cb, $hdr) = @_;
		711
		712	&_destroy_state ($state);
		713
		714	$cb->(undef, $hdr);
		715	()
		716	}
		717
		718	our %IDEMPOTENT = (
		719	DELETE => 1,
		720	GET => 1,
		721	QUERY => 1,
		722	HEAD => 1,
		723	OPTIONS => 1,
		724	PUT => 1,
		725	TRACE => 1,
		726
		727	ACL => 1,
		728	"BASELINE-CONTROL" => 1,
		729	BIND => 1,
		730	CHECKIN => 1,
		731	CHECKOUT => 1,
		732	COPY => 1,
		733	LABEL => 1,
		734	LINK => 1,
		735	MERGE => 1,
		736	MKACTIVITY => 1,
		737	MKCALENDAR => 1,
		738	MKCOL => 1,
		739	MKREDIRECTREF => 1,
		740	MKWORKSPACE => 1,
		741	MOVE => 1,
		742	ORDERPATCH => 1,
		743	PROPFIND => 1,
		744	PROPPATCH => 1,
		745	REBIND => 1,
		746	REPORT => 1,
		747	SEARCH => 1,
		748	UNBIND => 1,
		749	UNCHECKOUT => 1,
		750	UNLINK => 1,
		751	UNLOCK => 1,
		752	UPDATE => 1,
		753	UPDATEREDIRECTREF => 1,
		754	"VERSION-CONTROL" => 1,
		755	);
		756
108	sub http_request($$$;@) {	757	sub http_request($$@) {
109	my $cb = pop;	758	my $cb = pop;
110	my ($method, $url, %arg) = @_;	759	my ($method, $url, %arg) = @_;
111		760
112	my %hdr;	761	my %hdr;
113		762
		763	$arg{tls_ctx} = $TLS_CTX_LOW if $arg{tls_ctx} eq "low" || !exists $arg{tls_ctx};
		764	$arg{tls_ctx} = $TLS_CTX_HIGH if $arg{tls_ctx} eq "high";
		765
		766	$method = uc $method;
		767
114	if (my $hdr = delete $arg{headers}) {	768	if (my $hdr = $arg{headers}) {
115	while (my ($k, $v) = each %$hdr) {	769	while (my ($k, $v) = each %$hdr) {
116	$hdr{lc $k} = $v;	770	$hdr{lc $k} = $v;
117	}	771	}
118	}	772	}
119		773
120	my $proxy = $arg{proxy} || $PROXY;	774	# pseudo headers for all subsequent responses
		775	my @pseudo = (URL => $url);
		776	push @pseudo, Redirect => delete $arg{Redirect} if exists $arg{Redirect};
		777
		778	my $recurse = exists $arg{recurse} ? delete $arg{recurse} : $MAX_RECURSE;
		779
		780	return $cb->(undef, { @pseudo, Status => 599, Reason => "Too many redirections" })
		781	if $recurse < 0;
		782
		783	my $proxy = exists $arg{proxy} ? $arg{proxy} : $PROXY;
121	my $timeout = $arg{timeout} || $TIMEOUT;	784	my $timeout = $arg{timeout} || $TIMEOUT;
122		785
123	$hdr{"user-agent"} ||= $USERAGENT;	786	my ($uscheme, $uauthority, $upath, $query, undef) = # ignore fragment
		787	$url =~ m|^([^:]+):(?://([^/?#]*))?([^?#]*)(?:(\?[^#]*))?(?:#(.*))?$|;
124		788
125	my ($host, $port, $path, $scheme);	789	$uscheme = lc $uscheme;
		790
		791	my $uport = $uscheme eq "http" ? 80
		792	: $uscheme eq "https" ? 443
		793	: return $cb->(undef, { @pseudo, Status => 599, Reason => "Only http and https URL schemes supported" });
		794
		795	$uauthority =~ /^(?: .*\@ )? ([^\@]+?) (?: : (\d+) )?$/x
		796	or return $cb->(undef, { @pseudo, Status => 599, Reason => "Unparsable URL" });
		797
		798	my $uhost = lc $1;
		799	$uport = $2 if defined $2;
		800
		801	$hdr{host} = defined $2 ? "$uhost:$2" : "$uhost"
		802	unless exists $hdr{host};
		803
		804	$uhost =~ s/^\[(.*)\]$/$1/;
		805	$upath .= $query if length $query;
		806
		807	$upath =~ s%^/?%/%;
		808
		809	# cookie processing
		810	if (my $jar = $arg{cookie_jar}) {
		811	my $cookies = cookie_jar_extract $jar, $uscheme, $uhost, $upath;
		812
		813	$hdr{cookie} = join "; ", @$cookies
		814	if @$cookies;
		815	}
		816
		817	my ($rhost, $rport, $rscheme, $rpath); # request host, port, path
126		818
127	if ($proxy) {	819	if ($proxy) {
128	($host, $port, $scheme) = @$proxy;	820	($rpath, $rhost, $rport, $rscheme) = ($url, @$proxy);
129	$path = $url;	821
		822	$rscheme = "http" unless defined $rscheme;
		823
		824	# don't support https requests over https-proxy transport,
		825	# can't be done with tls as spec'ed, unless you double-encrypt.
		826	$rscheme = "http" if $uscheme eq "https" && $rscheme eq "https";
		827
		828	$rhost = lc $rhost;
		829	$rscheme = lc $rscheme;
130	} else {	830	} else {
131	($scheme, my $authority, $path, my $query, my $fragment) =	831	($rhost, $rport, $rscheme, $rpath) = ($uhost, $uport, $uscheme, $upath);
132	$url =~ m|(?:([^:/?#]+):)?(?://([^/?#]*))?([^?#]*)(?:\?([^#]*))?(?:#(.*))?|;
133
134	$port = $scheme eq "http" ? 80
135	: $scheme eq "https" ? 443
136	: croak "$url: only http and https URLs supported";
137
138	$authority =~ /^(?: .*\@ )? ([^\@:]+) (?: : (\d+) )?$/x
139	or croak "$authority: unparsable URL";
140
141	$host = $1;
142	$port = $2 if defined $2;
143
144	$host =~ s/^\[(.*)\]$/$1/;
145	$path .= "?$query" if length $query;
146
147	$path = "/" unless $path;
148
149	$hdr{host} = $host = lc $host;
150	}	832	}
151		833
152	$scheme = lc $scheme;	834	# leave out fragment and query string, just a heuristic
		835	$hdr{referer} = "$uscheme://$uauthority$upath" unless exists $hdr{referer};
		836	$hdr{"user-agent"} = $USERAGENT unless exists $hdr{"user-agent"};
153		837
154	my %state;
155
156	my $body = "";
157	$state{body} = $body;
158
159	$hdr{"content-length"} = length $body;	838	$hdr{"content-length"} = length $arg{body}
		839	if length $arg{body} || $method ne "GET";
160		840
161	$state{connect_guard} = AnyEvent::Socket::tcp_connect $host, $port, sub {	841	my $idempotent = $IDEMPOTENT{$method};
		842
		843	# default value for keepalive is true iff the request is for an idempotent method
		844	my $persistent = exists $arg{persistent} ? !!$arg{persistent} : $idempotent;
		845	my $keepalive = exists $arg{keepalive} ? !!$arg{keepalive} : !$proxy;
		846	my $was_persistent; # true if this is actually a recycled connection
		847
		848	# the key to use in the keepalive cache
		849	my $ka_key = "$uscheme\x00$uhost\x00$uport\x00$arg{sessionid}";
		850
		851	$hdr{connection} = ($persistent ? $keepalive ? "keep-alive, " : "" : "close, ") . "Te"; #1.1
		852	$hdr{te} = "trailers" unless exists $hdr{te}; #1.1
		853
		854	my %state = (connect_guard => 1);
		855
		856	my $ae_error = 595; # connecting
		857
		858	# handle actual, non-tunneled, request
		859	my $handle_actual_request = sub {
		860	$ae_error = 596; # request phase
		861
		862	my $hdl = $state{handle};
		863
		864	$hdl->starttls ("connect") if $uscheme eq "https" && !exists $hdl->{tls};
		865
		866	# send request
		867	$hdl->push_write (
		868	"$method $rpath HTTP/1.1\015\012"
		869	. (join "", map "\u$_: $hdr{$_}\015\012", grep defined $hdr{$_}, keys %hdr)
		870	. "\015\012"
		871	. $arg{body}
		872	);
		873
		874	# return if error occurred during push_write()
		875	return unless %state;
		876
		877	# reduce memory usage, save a kitten, also re-use it for the response headers.
		878	%hdr = ();
		879
		880	# status line and headers
		881	$state{read_response} = sub {
		882	return unless %state;
		883
		884	for ("$_[1]") {
		885	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
		886
		887	/^HTTP\/0*([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\012]*) )? \012/gxci
		888	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Invalid server response" };
		889
		890	# 100 Continue handling
		891	# should not happen as we don't send expect: 100-continue,
		892	# but we handle it just in case.
		893	# since we send the request body regardless, if we get an error
		894	# we are out of-sync, which we currently do NOT handle correctly.
		895	return $state{handle}->push_read (line => $qr_nlnl, $state{read_response})
		896	if $2 eq 100;
		897
		898	push @pseudo,
		899	HTTPVersion => $1,
		900	Status => $2,
		901	Reason => $3,
		902	;
		903
		904	my $hdr = _parse_hdr
		905	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Garbled response headers" };
		906
		907	%hdr = (%$hdr, @pseudo);
		908	}
		909
		910	# redirect handling
		911	# relative uri handling forced by microsoft and other shitheads.
		912	# we give our best and fall back to URI if available.
		913	if (exists $hdr{location}) {
		914	my $loc = $hdr{location};
		915
		916	if ($loc =~ m%^//%) { # //
		917	$loc = "$uscheme:$loc";
		918
		919	} elsif ($loc eq "") {
		920	$loc = $url;
		921
		922	} elsif ($loc !~ /^(?: $ | [^:\/?\#]+ : )/x) { # anything "simple"
		923	$loc =~ s/^\.\/+//;
		924
		925	if ($loc !~ m%^[.?#]%) {
		926	my $prefix = "$uscheme://$uauthority";
		927
		928	unless ($loc =~ s/^\///) {
		929	$prefix .= $upath;
		930	$prefix =~ s/\/[^\/]*$//;
		931	}
		932
		933	$loc = "$prefix/$loc";
		934
		935	} elsif (eval { require URI }) { # uri
		936	$loc = URI->new_abs ($loc, $url)->as_string;
		937
		938	} else {
		939	return _error %state, $cb, { @pseudo, Status => 599, Reason => "Cannot parse Location (URI module missing)" };
		940	#$hdr{Status} = 599;
		941	#$hdr{Reason} = "Unparsable Redirect (URI module missing)";
		942	#$recurse = 0;
		943	}
		944	}
		945
		946	$hdr{location} = $loc;
		947	}
		948
		949	my $redirect;
		950
		951	if ($recurse) {
		952	my $status = $hdr{Status};
		953
		954	# industry standard is to redirect POST as GET for
		955	# 301, 302 and 303, in contrast to HTTP/1.0 and 1.1.
		956	# also, the UA should ask the user for 301 and 307 and POST,
		957	# industry standard seems to be to simply follow.
		958	# we go with the industry standard. 308 is defined
		959	# by rfc7538
		960	if ($status == 301 or $status == 302 or $status == 303) {
		961	$redirect = 1;
		962	# HTTP/1.1 is unclear on how to mutate the method
		963	unless ($method eq "HEAD") {
		964	$method = "GET";
		965	delete $arg{body};
		966	}
		967	} elsif ($status == 307 or $status == 308) {
		968	$redirect = 1;
		969	}
		970	}
		971
		972	my $finish = sub { # ($data, $err_status, $err_reason[, $persistent])
		973	if ($state{handle}) {
		974	# handle keepalive
		975	if (
		976	$persistent
		977	&& $_[3]
		978	&& ($hdr{HTTPVersion} < 1.1
		979	? $hdr{connection} =~ /\bkeep-?alive\b/i
		980	: $hdr{connection} !~ /\bclose\b/i)
		981	) {
		982	ka_store $ka_key, delete $state{handle};
		983	} else {
		984	# no keepalive, destroy the handle
		985	$state{handle}->destroy;
		986	}
		987	}
		988
		989	%state = ();
		990
		991	if (defined $_[1]) {
		992	$hdr{OrigStatus} = $hdr{Status}; $hdr{Status} = $_[1];
		993	$hdr{OrigReason} = $hdr{Reason}; $hdr{Reason} = $_[2];
		994	}
		995
		996	# set-cookie processing
		997	if ($arg{cookie_jar}) {
		998	cookie_jar_set_cookie $arg{cookie_jar}, $hdr{"set-cookie"}, $uhost, $hdr{date};
		999	}
		1000
		1001	if ($redirect && exists $hdr{location}) {
		1002	# we ignore any errors, as it is very common to receive
		1003	# Content-Length != 0 but no actual body
		1004	# we also access %hdr, as $_[1] might be an erro
		1005	$state{recurse} =
		1006	http_request (
		1007	$method => $hdr{location},
		1008	%arg,
		1009	recurse => $recurse - 1,
		1010	Redirect => [$_[0], \%hdr],
		1011	sub {
		1012	%state = ();
		1013	&$cb
		1014	},
		1015	);
		1016	} else {
		1017	$cb->($_[0], \%hdr);
		1018	}
		1019	};
		1020
		1021	$ae_error = 597; # body phase
		1022
		1023	my $chunked = $hdr{"transfer-encoding"} =~ /\bchunked\b/i; # not quite correct...
		1024
		1025	my $len = $chunked ? undef : $hdr{"content-length"};
		1026
		1027	# body handling, many different code paths
		1028	# - no body expected
		1029	# - want_body_handle
		1030	# - te chunked
		1031	# - 2x length known (with or without on_body)
		1032	# - 2x length not known (with or without on_body)
		1033	if (!$redirect && $arg{on_header} && !$arg{on_header}(\%hdr)) {
		1034	$finish->(undef, 598 => "Request cancelled by on_header");
		1035	} elsif (
		1036	$hdr{Status} =~ /^(?:1..|204|205|304)$/
		1037	or $method eq "HEAD"
		1038	or (defined $len && $len == 0) # == 0, not !, because "0 " is true
		1039	) {
		1040	# no body
		1041	$finish->("", undef, undef, 1);
		1042
		1043	} elsif (!$redirect && $arg{want_body_handle}) {
		1044	$_[0]->on_eof (undef);
		1045	$_[0]->on_error (undef);
		1046	$_[0]->on_read (undef);
		1047
		1048	$finish->(delete $state{handle});
		1049
		1050	} elsif ($chunked) {
		1051	my $cl = 0;
		1052	my $body = "";
		1053	my $on_body = (!$redirect && $arg{on_body}) || sub { $body .= shift; 1 };
		1054
		1055	$state{read_chunk} = sub {
		1056	$_[1] =~ /^([0-9a-fA-F]+)/
		1057	or return $finish->(undef, $ae_error => "Garbled chunked transfer encoding");
		1058
		1059	my $len = hex $1;
		1060
		1061	if ($len) {
		1062	$cl += $len;
		1063
		1064	$_[0]->push_read (chunk => $len, sub {
		1065	$on_body->($_[1], \%hdr)
		1066	or return $finish->(undef, 598 => "Request cancelled by on_body");
		1067
		1068	$_[0]->push_read (line => sub {
		1069	length $_[1]
		1070	and return $finish->(undef, $ae_error => "Garbled chunked transfer encoding");
		1071	$_[0]->push_read (line => $state{read_chunk});
		1072	});
		1073	});
		1074	} else {
		1075	$hdr{"content-length"} ||= $cl;
		1076
		1077	$_[0]->push_read (line => $qr_nlnl, sub {
		1078	if (length $_[1]) {
		1079	for ("$_[1]") {
		1080	y/\015//d; # weed out any \015, as they show up in the weirdest of places.
		1081
		1082	my $hdr = _parse_hdr
		1083	or return $finish->(undef, $ae_error => "Garbled response trailers");
		1084
		1085	%hdr = (%hdr, %$hdr);
		1086	}
		1087	}
		1088
		1089	$finish->($body, undef, undef, 1);
		1090	});
		1091	}
		1092	};
		1093
		1094	$_[0]->push_read (line => $state{read_chunk});
		1095
		1096	} elsif (!$redirect && $arg{on_body}) {
		1097	if (defined $len) {
		1098	$_[0]->on_read (sub {
		1099	$len -= length $_[0]{rbuf};
		1100
		1101	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
		1102	or return $finish->(undef, 598 => "Request cancelled by on_body");
		1103
		1104	$len > 0
		1105	or $finish->("", undef, undef, 1);
		1106	});
		1107	} else {
		1108	$_[0]->on_eof (sub {
		1109	$finish->("");
		1110	});
		1111	$_[0]->on_read (sub {
		1112	$arg{on_body}(delete $_[0]{rbuf}, \%hdr)
		1113	or $finish->(undef, 598 => "Request cancelled by on_body");
		1114	});
		1115	}
		1116	} else {
		1117	$_[0]->on_eof (undef);
		1118
		1119	if (defined $len) {
		1120	$_[0]->on_read (sub {
		1121	$finish->((substr delete $_[0]{rbuf}, 0, $len, ""), undef, undef, 1)
		1122	if $len <= length $_[0]{rbuf};
		1123	});
		1124	} else {
		1125	$_[0]->on_error (sub {
		1126	($! == Errno::EPIPE || !$!)
		1127	? $finish->(delete $_[0]{rbuf})
		1128	: $finish->(undef, $ae_error => $_[2]);
		1129	});
		1130	$_[0]->on_read (sub { });
		1131	}
		1132	}
		1133	};
		1134
		1135	# if keepalive is enabled, then the server closing the connection
		1136	# before a response can happen legally - we retry on idempotent methods.
		1137	if ($was_persistent && $idempotent) {
		1138	my $old_eof = $hdl->{on_eof};
		1139	$hdl->{on_eof} = sub {
		1140	_destroy_state %state;
		1141
		1142	%state = ();
		1143	$state{recurse} =
		1144	http_request (
		1145	$method => $url,
		1146	%arg,
		1147	recurse => $recurse - 1,
		1148	persistent => 0,
		1149	sub {
		1150	%state = ();
		1151	&$cb
		1152	}
		1153	);
		1154	};
		1155	$hdl->on_read (sub {
		1156	return unless %state;
		1157
		1158	# as soon as we receive something, a connection close
		1159	# once more becomes a hard error
		1160	$hdl->{on_eof} = $old_eof;
		1161	$hdl->push_read (line => $qr_nlnl, $state{read_response});
		1162	});
		1163	} else {
		1164	$hdl->push_read (line => $qr_nlnl, $state{read_response});
		1165	}
		1166	};
		1167
		1168	my $prepare_handle = sub {
		1169	my ($hdl) = $state{handle};
		1170
		1171	$hdl->on_error (sub {
		1172	_error %state, $cb, { @pseudo, Status => $ae_error, Reason => $_[2] };
		1173	});
		1174	$hdl->on_eof (sub {
		1175	_error %state, $cb, { @pseudo, Status => $ae_error, Reason => "Unexpected end-of-file" };
		1176	});
		1177	$hdl->timeout_reset;
		1178	$hdl->timeout ($timeout);
		1179	};
		1180
		1181	# connected to proxy (or origin server)
		1182	my $connect_cb = sub {
162	$state{fh} = shift	1183	my $fh = shift
163	or return $cb->(undef, { Status => 599, Reason => "$!" });	1184	or return _error %state, $cb, { @pseudo, Status => $ae_error, Reason => "$!" };
164		1185
165	delete $state{connect_guard}; # reduce memory usage, save a tree	1186	return unless delete $state{connect_guard};
166		1187
167	# get handle	1188	# get handle
168	$state{handle} = new AnyEvent::Handle	1189	$state{handle} = new AnyEvent::Handle
169	fh => $state{fh},	1190	%{ $arg{handle_params} },
170	($scheme eq "https" ? (tls => "connect") : ());	1191	fh => $fh,
171		1192	peername => $uhost,
172	# limit the number of persistent connections	1193	tls_ctx => $arg{tls_ctx},
173	if ($KA_COUNT{$_[1]} < $MAX_PERSISTENT_PER_HOST) {
174	++$KA_COUNT{$_[1]};
175	$state{handle}{ka_count_guard} = AnyEvent::Util::guard { --$KA_COUNT{$_[1]} };
176	$hdr{connection} = "keep-alive";
177	delete $hdr{connection}; # keep-alive not yet supported
178	} else {
179	delete $hdr{connection};
180	}
181
182	# (re-)configure handle
183	$state{handle}->timeout ($timeout);
184	$state{handle}->on_error (sub {
185	%state = ();
186	$cb->(undef, { Status => 599, Reason => "$!" });
187	});
188	$state{handle}->on_eof (sub {
189	%state = ();
190	$cb->(undef, { Status => 599, Reason => "unexpected end-of-file" });
191	});
192
193	# send request
194	$state{handle}->push_write (
195	"\U$method\E $path HTTP/1.0\015\012"
196	. (join "", map "$_: $hdr{$_}\015\012", keys %hdr)
197	. "\015\012"
198	. (delete $state{body})
199	);	1194	;
200		1195
201	%hdr = (); # reduce memory usage, save a kitten	1196	$prepare_handle->();
202		1197
203	# status line	1198	#$state{handle}->starttls ("connect") if $rscheme eq "https";
		1199
		1200	# now handle proxy-CONNECT method
		1201	if ($proxy && $uscheme eq "https") {
		1202	# oh dear, we have to wrap it into a connect request
		1203
		1204	my $auth = exists $hdr{"proxy-authorization"}
		1205	? "proxy-authorization: " . (delete $hdr{"proxy-authorization"}) . "\015\012"
		1206	: "";
		1207
		1208	# maybe re-use $uauthority with patched port?
		1209	$state{handle}->push_write ("CONNECT $uhost:$uport HTTP/1.0\015\012$auth\015\012");
204	$state{handle}->push_read (line => qr/\015?\012/, sub {	1210	$state{handle}->push_read (line => $qr_nlnl, sub {
205	$_[1] =~ /^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) \s+ ([^\015\012]+)/ix	1211	$_[1] =~ /^HTTP\/([0-9\.]+) \s+ ([0-9]{3}) (?: \s+ ([^\015\012]*) )?/ix
206	or return (%state = (), $cb->(undef, { Status => 599, Reason => "invalid server response ($_[1])" }));	1212	or return _error %state, $cb, { @pseudo, Status => 599, Reason => "Invalid proxy connect response ($_[1])" };
207		1213
208	my %hdr = ( # response headers	1214	if ($2 == 200) {
209	HTTPVersion => ",$1",	1215	$rpath = $upath;
210	Status => ",$2",	1216	$handle_actual_request->();
211	Reason => ",$3",
212	);
213
214	# headers, could be optimized a bit
215	$state{handle}->unshift_read (line => qr/\015?\012\015?\012/, sub {
216	for ("$_[1]\012") {
217	# we support spaces in field names, as lotus domino
218	# creates them.
219	$hdr{lc $1} .= ",$2"
220	while /\G
221	([^:\000-\037]+):
222	[\011\040]*
223	((?: [^\015\012]+ | \015?\012[\011\040] )*)
224	\015?\012
225	/gxc;
226
227	/\G$/
228	or return $cb->(undef, { Status => 599, Reason => "garbled response headers" });
229	}
230
231	substr $_, 0, 1, ""
232	for values %hdr;
233
234	if (exists $hdr{"content-length"}) {
235	$_[0]->unshift_read (chunk => $hdr{"content-length"}, sub {
236	# could cache persistent connection now
237	if ($hdr{connection} =~ /\bkeep-alive\b/i) {
238	# but we don't, due to misdesigns, this is annoyingly complex
239	};
240
241	%state = ();
242	$cb->($_[1], \%hdr);
243	});
244	} else {	1217	} else {
245	# too bad, need to read until we get an error or EOF,	1218	_error %state, $cb, { @pseudo, Status => $2, Reason => $3 };
246	# no way to detect winged data.
247	$_[0]->on_error (sub {
248	%state = ();
249	$cb->($_[0]{rbuf}, \%hdr);
250	});
251	$_[0]->on_eof (undef);
252	$_[0]->on_read (sub { });
253	}	1219	}
254	});	1220	});
		1221	} else {
		1222	delete $hdr{"proxy-authorization"} unless $proxy;
		1223
		1224	$handle_actual_request->();
255	});	1225	}
256	}, sub {
257	$timeout
258	};	1226	};
259		1227
		1228	_get_slot $uhost, sub {
		1229	$state{slot_guard} = shift;
		1230
		1231	return unless $state{connect_guard};
		1232
		1233	# try to use an existing keepalive connection, but only if we, ourselves, plan
		1234	# on a keepalive request (in theory, this should be a separate config option).
		1235	if ($persistent && $KA_CACHE{$ka_key}) {
		1236	$was_persistent = 1;
		1237
		1238	$state{handle} = ka_fetch $ka_key;
		1239	# $state{handle}->destroyed
		1240	# and die "AnyEvent::HTTP: unexpectedly got a destructed handle (1), please report.";#d#
		1241	$prepare_handle->();
		1242	# $state{handle}->destroyed
		1243	# and die "AnyEvent::HTTP: unexpectedly got a destructed handle (2), please report.";#d#
		1244	$rpath = $upath;
		1245	$handle_actual_request->();
		1246
		1247	} else {
		1248	my $tcp_connect = $arg{tcp_connect}
		1249	|| do { require AnyEvent::Socket; \&AnyEvent::Socket::tcp_connect };
		1250
		1251	$state{connect_guard} = $tcp_connect->($rhost, $rport, $connect_cb, $arg{on_prepare} || sub { $timeout });
		1252	}
		1253	};
		1254
260	defined wantarray && AnyEvent::Util::guard { %state = () }	1255	defined wantarray && AnyEvent::Util::guard { _destroy_state %state }
261	}	1256	}
262		1257
263	sub http_get($$;@) {	1258	sub http_get($@) {
264	unshift @_, "GET";	1259	unshift @_, "GET";
265	&http_request	1260	&http_request
266	}	1261	}
267		1262
		1263	sub http_head($@) {
		1264	unshift @_, "HEAD";
		1265	&http_request
		1266	}
		1267
		1268	sub http_post($$@) {
		1269	my $url = shift;
		1270	unshift @_, "POST", $url, "body";
		1271	&http_request
		1272	}
		1273
		1274	=back
		1275
		1276	=head2 DNS CACHING
		1277
		1278	AnyEvent::HTTP uses the AnyEvent::Socket::tcp_connect function for
		1279	the actual connection, which in turn uses AnyEvent::DNS to resolve
		1280	hostnames. The latter is a simple stub resolver and does no caching
		1281	on its own. If you want DNS caching, you currently have to provide
		1282	your own default resolver (by storing a suitable resolver object in
		1283	C<$AnyEvent::DNS::RESOLVER>) or your own C<tcp_connect> callback.
		1284
268	=head2 GLOBAL FUNCTIONS AND VARIABLES	1285	=head2 GLOBAL FUNCTIONS AND VARIABLES
269		1286
270	=over 4	1287	=over 4
271		1288
272	=item AnyEvent::HTTP::set_proxy "proxy-url"	1289	=item AnyEvent::HTTP::set_proxy "proxy-url"
273		1290
274	Sets the default proxy server to use. The proxy-url must begin with a	1291	Sets the default proxy server to use. The proxy-url must begin with a
275	string of the form C<http://host:port> (optionally C<https:...>).	1292	string of the form C<http://host:port>, croaks otherwise.
276		1293
		1294	To clear an already-set proxy, use C<undef>.
		1295
		1296	When AnyEvent::HTTP is loaded for the first time it will query the
		1297	default proxy from the operating system, currently by looking at
		1298	C<$ENV{http_proxy>}.
		1299
		1300	=item AnyEvent::HTTP::cookie_jar_expire $jar[, $session_end]
		1301
		1302	Remove all cookies from the cookie jar that have been expired. If
		1303	C<$session_end> is given and true, then additionally remove all session
		1304	cookies.
		1305
		1306	You should call this function (with a true C<$session_end>) before you
		1307	save cookies to disk, and you should call this function after loading them
		1308	again. If you have a long-running program you can additionally call this
		1309	function from time to time.
		1310
		1311	A cookie jar is initially an empty hash-reference that is managed by this
		1312	module. Its format is subject to change, but currently it is as follows:
		1313
		1314	The key C<version> has to contain C<2>, otherwise the hash gets
		1315	cleared. All other keys are hostnames or IP addresses pointing to
		1316	hash-references. The key for these inner hash references is the
		1317	server path for which this cookie is meant, and the values are again
		1318	hash-references. Each key of those hash-references is a cookie name, and
		1319	the value, you guessed it, is another hash-reference, this time with the
		1320	key-value pairs from the cookie, except for C<expires> and C<max-age>,
		1321	which have been replaced by a C<_expires> key that contains the cookie
		1322	expiry timestamp. Session cookies are indicated by not having an
		1323	C<_expires> key.
		1324
		1325	Here is an example of a cookie jar with a single cookie, so you have a
		1326	chance of understanding the above paragraph:
		1327
		1328	{
		1329	version => 2,
		1330	"10.0.0.1" => {
		1331	"/" => {
		1332	"mythweb_id" => {
		1333	_expires => 1293917923,
		1334	value => "ooRung9dThee3ooyXooM1Ohm",
		1335	},
		1336	},
		1337	},
		1338	}
		1339
		1340	=item $date = AnyEvent::HTTP::format_date $timestamp
		1341
		1342	Takes a POSIX timestamp (seconds since the epoch) and formats it as a HTTP
		1343	Date (RFC 2616).
		1344
		1345	=item $timestamp = AnyEvent::HTTP::parse_date $date
		1346
		1347	Takes a HTTP Date (RFC 2616) or a Cookie date (netscape cookie spec) or a
		1348	bunch of minor variations of those, and returns the corresponding POSIX
		1349	timestamp, or C<undef> if the date cannot be parsed.
		1350
277	=item $AnyEvent::HTTP::MAX_REDIRECTS	1351	=item $AnyEvent::HTTP::MAX_RECURSE
278		1352
279	The default value for the C<max_redirects> request parameter	1353	The default value for the C<recurse> request parameter (default: C<10>).
280	(default: C<10>).	1354
		1355	=item $AnyEvent::HTTP::TIMEOUT
		1356
		1357	The default timeout for connection operations (default: C<300>).
281		1358
282	=item $AnyEvent::HTTP::USERAGENT	1359	=item $AnyEvent::HTTP::USERAGENT
283		1360
284	The default value for the C<User-Agent> header (the default is	1361	The default value for the C<User-Agent> header (the default is
285	C<Mozilla/5.0 (compatible; AnyEvent::HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)>).	1362	C<Mozilla/5.0 (compatible; U; AnyEvent-HTTP/$VERSION; +http://software.schmorp.de/pkg/AnyEvent)>).
286		1363
287	=item $AnyEvent::HTTP::MAX_PERSISTENT	1364	=item $AnyEvent::HTTP::MAX_PER_HOST
288		1365
289	The maximum number of persistent connections to keep open (default: 8).	1366	The maximum number of concurrent connections to the same host (identified
		1367	by the hostname). If the limit is exceeded, then additional requests
		1368	are queued until previous connections are closed. Both persistent and
		1369	non-persistent connections are counted in this limit.
		1370
		1371	The default value for this is C<4>, and it is highly advisable to not
		1372	increase it much.
		1373
		1374	For comparison: the RFC's recommend 4 non-persistent or 2 persistent
		1375	connections, older browsers used 2, newer ones (such as firefox 3)
		1376	typically use 6, and Opera uses 8 because like, they have the fastest
		1377	browser and give a shit for everybody else on the planet.
290		1378
291	=item $AnyEvent::HTTP::PERSISTENT_TIMEOUT	1379	=item $AnyEvent::HTTP::PERSISTENT_TIMEOUT
292		1380
293	The maximum time to cache a persistent connection, in seconds (default: 2).	1381	The time after which idle persistent connections get closed by
		1382	AnyEvent::HTTP (default: C<3>).
		1383
		1384	=item $AnyEvent::HTTP::ACTIVE
		1385
		1386	The number of active connections. This is not the number of currently
		1387	running requests, but the number of currently open and non-idle TCP
		1388	connections. This number can be useful for load-leveling.
294		1389
295	=back	1390	=back
296		1391
297	=cut	1392	=cut
298		1393
		1394	our @month = qw(Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec);
		1395	our @weekday = qw(Sun Mon Tue Wed Thu Fri Sat);
		1396
		1397	sub format_date($) {
		1398	my ($time) = @_;
		1399
		1400	# RFC 822/1123 format
		1401	my ($S, $M, $H, $mday, $mon, $year, $wday, $yday, undef) = gmtime $time;
		1402
		1403	sprintf "%s, %02d %s %04d %02d:%02d:%02d GMT",
		1404	$weekday[$wday], $mday, $month[$mon], $year + 1900,
		1405	$H, $M, $S;
		1406	}
		1407
		1408	sub parse_date($) {
		1409	my ($date) = @_;
		1410
		1411	my ($d, $m, $y, $H, $M, $S);
		1412
		1413	if ($date =~ /^[A-Z][a-z][a-z]+, ([0-9][0-9]?)[\- ]([A-Z][a-z][a-z])[\- ]([0-9][0-9][0-9][0-9]) ([0-9][0-9]?):([0-9][0-9]?):([0-9][0-9]?) GMT$/) {
		1414	# RFC 822/1123, required by RFC 2616 (with " ")
		1415	# cookie dates (with "-")
		1416
		1417	($d, $m, $y, $H, $M, $S) = ($1, $2, $3, $4, $5, $6);
		1418
		1419	} elsif ($date =~ /^[A-Z][a-z][a-z]+, ([0-9][0-9]?)-([A-Z][a-z][a-z])-([0-9][0-9]) ([0-9][0-9]?):([0-9][0-9]?):([0-9][0-9]?) GMT$/) {
		1420	# RFC 850
		1421	($d, $m, $y, $H, $M, $S) = ($1, $2, $3 < 69 ? $3 + 2000 : $3 + 1900, $4, $5, $6);
		1422
		1423	} elsif ($date =~ /^[A-Z][a-z][a-z]+ ([A-Z][a-z][a-z]) ([0-9 ]?[0-9]) ([0-9][0-9]?):([0-9][0-9]?):([0-9][0-9]?) ([0-9][0-9][0-9][0-9])$/) {
		1424	# ISO C's asctime
		1425	($d, $m, $y, $H, $M, $S) = ($2, $1, $6, $3, $4, $5);
		1426	}
		1427	# other formats fail in the loop below
		1428
		1429	for (0..11) {
		1430	if ($m eq $month[$_]) {
		1431	require Time::Local;
		1432	return eval { Time::Local::timegm ($S, $M, $H, $d, $_, $y) };
		1433	}
		1434	}
		1435
		1436	undef
		1437	}
		1438
299	sub set_proxy($) {	1439	sub set_proxy($) {
300	$PROXY = [$2, $3 || 3128, $1] if $_[0] =~ m%^(https?):// ([^:/]+) (?: : (\d*) )?%ix;	1440	if (length $_[0]) {
		1441	$_[0] =~ m%^(http):// ([^:/]+) (?: : (\d*) )?%ix
		1442	or Carp::croak "$_[0]: invalid proxy URL";
		1443	$PROXY = [$2, $3 || 3128, $1]
		1444	} else {
		1445	undef $PROXY;
		1446	}
301	}	1447	}
302		1448
303	# initialise proxy from environment	1449	# initialise proxy from environment
		1450	eval {
304	set_proxy $ENV{http_proxy};	1451	set_proxy $ENV{http_proxy};
		1452	};
		1453
		1454	=head2 SHOWCASE
		1455
		1456	This section contains some more elaborate "real-world" examples or code
		1457	snippets.
		1458
		1459	=head2 HTTP/1.1 FILE DOWNLOAD
		1460
		1461	Downloading files with HTTP can be quite tricky, especially when something
		1462	goes wrong and you want to resume.
		1463
		1464	Here is a function that initiates and resumes a download. It uses the
		1465	last modified time to check for file content changes, and works with many
		1466	HTTP/1.0 servers as well, and usually falls back to a complete re-download
		1467	on older servers.
		1468
		1469	It calls the completion callback with either C<undef>, which means a
		1470	nonretryable error occurred, C<0> when the download was partial and should
		1471	be retried, and C<1> if it was successful.
		1472
		1473	use AnyEvent::HTTP;
		1474
		1475	sub download($$$) {
		1476	my ($url, $file, $cb) = @_;
		1477
		1478	open my $fh, "+<", $file
		1479	or die "$file: $!";
		1480
		1481	my %hdr;
		1482	my $ofs = 0;
		1483
		1484	if (stat $fh and -s _) {
		1485	$ofs = -s _;
		1486	warn "-s is ", $ofs;
		1487	$hdr{"if-unmodified-since"} = AnyEvent::HTTP::format_date +(stat _)[9];
		1488	$hdr{"range"} = "bytes=$ofs-";
		1489	}
		1490
		1491	http_get $url,
		1492	headers => \%hdr,
		1493	on_header => sub {
		1494	my ($hdr) = @_;
		1495
		1496	if ($hdr->{Status} == 200 && $ofs) {
		1497	# resume failed
		1498	truncate $fh, $ofs = 0;
		1499	}
		1500
		1501	sysseek $fh, $ofs, 0;
		1502
		1503	1
		1504	},
		1505	on_body => sub {
		1506	my ($data, $hdr) = @_;
		1507
		1508	if ($hdr->{Status} =~ /^2/) {
		1509	length $data == syswrite $fh, $data
		1510	or return; # abort on write errors
		1511	}
		1512
		1513	1
		1514	},
		1515	sub {
		1516	my (undef, $hdr) = @_;
		1517
		1518	my $status = $hdr->{Status};
		1519
		1520	if (my $time = AnyEvent::HTTP::parse_date $hdr->{"last-modified"}) {
		1521	utime $time, $time, $fh;
		1522	}
		1523
		1524	if ($status == 200 || $status == 206 || $status == 416) {
		1525	# download ok || resume ok || file already fully downloaded
		1526	$cb->(1, $hdr);
		1527
		1528	} elsif ($status == 412) {
		1529	# file has changed while resuming, delete and retry
		1530	unlink $file;
		1531	$cb->(0, $hdr);
		1532
		1533	} elsif ($status == 500 or $status == 503 or $status =~ /^59/) {
		1534	# retry later
		1535	$cb->(0, $hdr);
		1536
		1537	} else {
		1538	$cb->(undef, $hdr);
		1539	}
		1540	}
		1541	;
		1542	}
		1543
		1544	download "http://server/somelargefile", "/tmp/somelargefile", sub {
		1545	if ($_[0]) {
		1546	print "OK!\n";
		1547	} elsif (defined $_[0]) {
		1548	print "please retry later\n";
		1549	} else {
		1550	print "ERROR\n";
		1551	}
		1552	};
		1553
		1554	=head3 SOCKS PROXIES
		1555
		1556	Socks proxies are not directly supported by AnyEvent::HTTP. You can
		1557	compile your perl to support socks, or use an external program such as
		1558	F<socksify> (dante) or F<tsocks> to make your program use a socks proxy
		1559	transparently.
		1560
		1561	Alternatively, for AnyEvent::HTTP only, you can use your own
		1562	C<tcp_connect> function that does the proxy handshake - here is an example
		1563	that works with socks4a proxies:
		1564
		1565	use Errno;
		1566	use AnyEvent::Util;
		1567	use AnyEvent::Socket;
		1568	use AnyEvent::Handle;
		1569
		1570	# host, port and username of/for your socks4a proxy
		1571	my $socks_host = "10.0.0.23";
		1572	my $socks_port = 9050;
		1573	my $socks_user = "";
		1574
		1575	sub socks4a_connect {
		1576	my ($host, $port, $connect_cb, $prepare_cb) = @_;
		1577
		1578	my $hdl = new AnyEvent::Handle
		1579	connect => [$socks_host, $socks_port],
		1580	on_prepare => sub { $prepare_cb->($_[0]{fh}) },
		1581	on_error => sub { $connect_cb->() },
		1582	;
		1583
		1584	$hdl->push_write (pack "CCnNZ*Z*", 4, 1, $port, 1, $socks_user, $host);
		1585
		1586	$hdl->push_read (chunk => 8, sub {
		1587	my ($hdl, $chunk) = @_;
		1588	my ($status, $port, $ipn) = unpack "xCna4", $chunk;
		1589
		1590	if ($status == 0x5a) {
		1591	$connect_cb->($hdl->{fh}, (format_address $ipn) . ":$port");
		1592	} else {
		1593	$! = Errno::ENXIO; $connect_cb->();
		1594	}
		1595	});
		1596
		1597	$hdl
		1598	}
		1599
		1600	Use C<socks4a_connect> instead of C<tcp_connect> when doing C<http_request>s,
		1601	possibly after switching off other proxy types:
		1602
		1603	AnyEvent::HTTP::set_proxy undef; # usually you do not want other proxies
		1604
		1605	http_get 'http://www.google.com', tcp_connect => \&socks4a_connect, sub {
		1606	my ($data, $headers) = @_;
		1607	...
		1608	};
305		1609
306	=head1 SEE ALSO	1610	=head1 SEE ALSO
307		1611
308	L<AnyEvent>.	1612	L<AnyEvent>.
309		1613
310	=head1 AUTHOR	1614	=head1 AUTHOR
311		1615
312	Marc Lehmann <schmorp@schmorp.de>	1616	Marc Lehmann <schmorp@schmorp.de>
313	http://home.schmorp.de/	1617	http://home.schmorp.de/
		1618
		1619	With many thanks to Дмитрий Шалашов, who provided countless
		1620	testcases and bugreports.
314		1621
315	=cut	1622	=cut
316		1623
317	1	1624	1
318		1625

Diff Legend

–	Removed lines
+	Added lines
<	Changed lines
>	Changed lines